Friday, July 13, 2012

Did Toxic Effects of an EHR Kill Rory Staunton?

A stunning story about the death of a young man from sepsis (blood poisoning by infection), missed in an ED, appeared in the New York Times:

An Infection, Unnoticed, Turns Unstoppable

By JIM DWYER
New York Times
Published: July 11, 2012

For a moment, an emergency room doctor stepped away from the scrum of people working on Rory Staunton, 12, and spoke to his parents.

“Your son is seriously ill,” the doctor said.

“How seriously?” Rory’s mother, Orlaith Staunton, asked.

The doctor paused.

“Gravely ill,” he said.

How could that be?

Two days earlier, diving for a basketball at his school gym, Rory had cut his arm. He arrived at his pediatrician’s office the next day, Thursday, March 29, vomiting, feverish and with pain in his leg. He was sent to the emergency room at NYU Langone Medical Center. The doctors agreed: He was suffering from an upset stomach and dehydration. He was given fluids, told to take Tylenol, and sent home.

Partially camouflaged by ordinary childhood woes, Rory’s condition was, in fact, already dire. Bacteria had gotten into his blood, probably through the cut on his arm. He was sliding into a septic crisis, an avalanche of immune responses to infection from which he would not escape. On April 1, three nights after he was sent home from the emergency room, he died in the intensive care unit. The cause was severe septic shock brought on by the infection, hospital records say. 

Rory Staunton, age 12, 5 feet 9 inches tall and 169 pounds, had suffered a cut on his arm.  He presented with a marked fever of 102 F (39 C), pulse markedly elevated at 131, respiratory rate elevated at 22; reported to have hit as high as 36 breaths per minute (in essence, panting).  It was reported by the NYT that before the ED visit his parents said his temperature had reached 104 F (40 C).

That alone should have set off some level of concern.  (It is possible narrative details of his history never made it into the ED chart; ED EHR's are often templated point-and-click affairs that can impair or discourage capture of narrative.)

Per the NYT, the bacteria Streptococcus pyogenes normally dwells in the throat or on the skin, areas where the body is well defended.  Also known as Group A streptococcus, the strain typically causes strep throat or impetigo.  However, if it gets into the blood stream (e.g., via a cut in the skin, as this patient suffered playing ball), the results can be devastating.

The lab results from the first ED visit are particularly stunning:


(From NYT article; click to enlarge)


The white blood cell count is markedly elevated at 14.7, meaning 14,700 cells per microliter of blood (cubic millimeter or 1 mm3).  Further, there is a plain evidence of greatly accelerated new white cell production, in the form of "bands", at 53% of the total (normally 5-15%).  Bands are immature white blood cells that are seen in the blood, being produced as part of the body's response to infection.

Herein is a significant issue.  The NYT noted that:

"Three hours later [i.e., after the ED visit, which reportedly only lasted 2 hours - ed.], Rory’s blood tests came back. High levels of neutrophils and “bands” – immature white blood cells – are evidence of infection. But nobody called the Stauntons, and by the time Rory returned to the hospital the next day, his infection was unstoppable. He died two days later."

Not getting into the issue regarding the patient apparently being discharged before the labs got back (itself an invitation to disaster), and the other abnormalities such as low sodium, low platelets, elevated glucose all pointing to a very sick patient... nobody called the Stantons with white cell results like these?   Nobody entertained the thought of ... antibiotics as a precaution?

It is possible - dare I say likely - that no clinical person in the ED ever saw these results.

EHR's that are poorly designed or implemented can have a toxic effect on care.  For instance, EHR's can cause user confusion if the user interface is complex, data can be lost due to poor relational design.  Data from the wrong patient's data can be presented (misidentification), or data from a lab can come back to the system after a patient has left, and despite being abnormal, just sit there in a silo without being looked at ("out of sight, out of mind"; a "silent silo" syndrome).

It is usually difficult to ascertain exactly which EHR product is being used at a particular hospital.  I note this medical center actively promoted its EPIC EHR in a June 2011 press release "NYU Langone Medical Center Launches Next Phase of Its Electronic Health Record System", although another system "ICIS"  (for Integrated Clinical Information System, "a state-of-the-art healthcare information management system that connects all NYULMC caregivers involved in patient care") is mentioned here.  The ICIS may also contain the Eclipsys Sunrise Clinical Manager, per this link. (I'd noted some clinically relevant problems with the latter in an FDA report here.)

In any case, magical powers are attributed to the technology that are not strongly or uniformly supported by the literature (link), but strongly pushed by industry marketing memes of deterministic health IT benefits and absolute beneficence:

“... Our electronic health record system is an integral part of our ongoing efforts to leverage technology and enhance our ability to provide patient-centered care and enable the highest level of quality care management,” said Bernard A. Birnbaum, MD, senior vice president and vice dean, chief of hospital operations at NYU Langone. “These front-end and back-end services are an important step in assuring our patient’s experience from beginning to end is a seamless one.”

I've documented examples of situations where EHR's and other IT components of clinical ERP systems (enterprise resource planning and management systems, a term that more accurately describes what exists in many hospitals now than the misleading, file cabinet-evoking term "EHR") contributed to or caused patient harm, such as at "Babies' deaths spotlight safety risks linked to computerized systems" - a computer error caused a central line placement x-ray to have gone unread, leading to death; "The Sweet death that wasn't very sweet" - a missing "difficult intubation" EHR flag led to a middle-aged man suffocating during an intubation attempt; and others.  An Australian researcher thoroughly studied the potential risks of an EHR meant specifically for ED's ("A Study of an Enterprise Health information System", PDF executive summary at this link).

The following passage in the NYT article also offers another clue:

... Two hours later, though, he had three [signs of sepsis]: his temperature had risen to 102, his pulse was 131 and his respiration rate was 22. But by the time those vital signs were recorded, at 9:26 p.m., they had no bearing on his treatment. In fact, the doctor had already decided that Rory was going home. Rory’s “ExitCare” instructions, signed by his father, were printed 12 minutes before those readings. 

Did those readings escape notice due to delayed charting (data entry), a common problem with EHRs in busy clinical environments?

The Institute of Medicine in its 2011 IOM report on health IT safety admitted harms are reported but the magnitude of harms is unknown due to multiple reporting impediments, as did the FDA in its 2010 internal memo on "H-IT Safety Issues" divulged by the Huffington Post Investigative Fund (see here and here).  The National Institute of Standards and Technology (NIST) admits in its 2011 report on HIT usability that EHR usability is often poor and may lead to "use error" (error caused or promoted by poor design, as distinguished from simple user error, see here), magnitude of problem also unknown.

In a startling medical situation such as Rory Stanton's, where crucial labs seem to have evaporated causing or contributing to delayed treatment of a devastating and obvious illness, I believe EHR-related factors need to be examined and ruled out first.

For, quite simply, if the EHR caused or contributed to this tragic debacle, the public could be at risk.


-- SS

Additional thought:  could this be the "cybernetic Libby Zion case" I've written of?

-- SS

July 18, 2012 Addendum:

The Stauntons, who appeared on the NBC Today Show are seeking to create a “Rory’s Law” in New York to ensure that parents have full access to blood and lab tests done on their children as soon as results are available, and that a doctor will be present to assess the findings. Story here.

-- SS

15 comments:

dmillho1 said...

While the EHR may have contributed to this sad event, I believe the person who collected the data just before this boy was discharged is accountable as well. That person should have seen that a change was happening and reported to someone who could and would take action.

Anonymous said...

I hope they sue the EMR vendor.

Anonymous said...

The fact that the hospital has a special "Stop Sepsis" program, with an associated "screening tool", makes this story even more remarkable. Could clinicians be relying too much on this tool to detect sepsis, and not be sufficiently aware of limitations it may have?

"A screening tool in the Stop Sepsis program, used when a patient first arrives in the emergency room, calls special attention to a person with three symptoms of a possible eight."

From the above description of the tool, I get the impression that its functional role is limited to patient data collected at the time of ED arrival.

If this tool were designed, and implemented, to detect and alert clinicians to 3 or more signs of sepsis recorded in the EHR, for any patient, at any time, it should have done so for Rory 2 hours after being admitted to the ED. The alert would have occurred on the new vitals signs data prior to availability of lab results.

dmillho1 said...

While the EHR may have contributed to this sad event, the person who took the last vital signs is culpable as well. Regardless of the discharge order already written, the finding of significant fever, tachycardia, etc. should have raised alarms for the staff person and sent that individual to someone who could make a clinical-care decision.

InformaticsMD said...

dmillho1 said...

While the EHR may have contributed to this sad event, the person who took the last vital signs is culpable as well.

The vital signs are arguable; the labs, I do not believe, are. Of course, there are multiple possible levels of responsibility possible - clinicians themselves, and corporate (if the IT contributed to a dangerous environment).

A root cause, however, may have been what I have termed "inappropriate overconfidence in computing."

The causality needs investigation, but the IT, its implementation, and its implementers should not be let off the hook if the "silent silo" effect was a factor in these labs never being acted on.

-- SS

Anonymous said...

Machines take vital signs and send them to the EHR. If a person is involved, such is an aid who puts them in the EHR. Nurses spend hours charting while the patients lie suffering.

The doctors will be blamed but the likely facilitator is the EHR that did not indicate new data had arrived and that served as an impediment to the workflow that would have otherwise raised the sepsis warnings.

Keith said...

Seems hard to blame the EMR in this awful event without more details. Most EMRs I have used require the doctor who ordered the test to lectronically sign off on the result. How this works in an emergenccy room where docs may be working shifts and and a doctor releasing a patient may not be the doctor that ordered the test may pose a problem.

Compare to the old method of paper reslts being sent to the ER, this is clearly an improvement, but poor set up could clearly contribute to a catastrophe.

Seems to me that this is more an issue of process where the medical personel locked themselves into a presumed diagnosis and did not wait to confirm the test results against their presumption before releasing the patient. Why would it take so long to get a CBC back from the lab as well? Seems more of a human error of not checking al the results prior to discharge.

Steve Lucas said...

One issue often pointed out is the world wide nature of this problem. Ontario, Canada is behind schedule and over budget on an ehealth project, but we find:

“Mitchell said the registry is now in testing with CGI Consultants Inc., the vendor that won the contract to build the registry in August 2010, and that any questions about why it was delayed or when it would be complete should be directed to the company.

That’s despite the fact the agreement between CGI and eHealth Ontario says eHealth has control over the disclosure of any information with respect to the project and that CGI can’t talk about it publicly without the agency’s approval.

“Under the CGI contract any media disclosures must be agreed to by CGI,” Sikin Samji-Clark said.”

Additionally there are issues with moving imaging and lab test between physicians and hospitals.

One has to wonder with the large reliance on technology and off site imaging studies how many more people will be injured by faulty IT?

We also have to wonder about the often placed gag orders covering the release of negative information regarding IT projects.

http://m.torontosun.com/2012/07/14/ehealth-needs-surgery

Steve Lucas

InformaticsMD said...

Steve Lucas wrote:

We also have to wonder about the often placed gag orders covering the release of negative information regarding IT projects.

The industry is spoiled. Through decades of misinformation and influence it has been given unprecedented special regulatory accommodation compared to other healthcare sectors, and to other risk-prone industries such as aviation, automotive, energy, etc., as I've written.

-- SS

InformaticsMD said...

"Keith" anonymous wrote:

Seems to me that this is more an issue of process where the medical personel locked themselves into a presumed diagnosis and did not wait to confirm the test results against their presumption before releasing the patient

Whatever happened was surely multi-factorial (that is, the dreaded "alignment of the holes in the cheese"). However, as I've written, EHR's toxic effects must be evaluated as a potential contributor or root cause. The "silent silo" syndrome is real - and inexcusable, as are the deleterious effects of impairment of narrative capture.

-- SS

Anonymous said...

Was there mayhem in that ER due to a crash of the EHR and/or other electronic and digital medical care infrastructure? Just wondering how otherwise competent doctors would have not acted on the constellation of data had they seen (or was it lurking in an electronic silo?) it and connected the dots. Hmmm.

InformaticsMD said...

Anonymous July 18, 2012 7:52:00 PM EDT wrote:

Was there mayhem in that ER due to a crash of the EHR and/or other electronic and digital medical care infrastructure?

That is indeed another consideration that should be investigated. See for instance "Twelve Hour Health IT Glitch at Allegheny General Hospital - But Patients Unaffected, Of Course" and other posts indexed under the term "glitch."

-- SS

Anonymous said...

You ask: "Did Toxic Effects of an EHR Kill Rory Staunton?'

Yes, until proven otherwise by the NYU leadership providing evidence to support their denial.

Anonymous said...

I believe in the comments the writer of the article references written nurses and doctor notes.

Anonymous said...

If this hospital is really was using the system called Epic as mentioned in the article, then I can tell you from personal experience, that there is no requirement or "hard stop" that forces you to check or sign off on labs or vital signs...It would be very easy to send someone home and then have VS or labs come back after the person has been discharged...